Consumer Driven Health Plans:
Employer Take-Up & Average Contributions
Justin R. Cress ∗
Abstract
In an era of rising health care costs, employers seek cost reductionstrategies, many relying on consumer driven health plans (CDHP). Uti-lizing data from a nationally representative survey of firms, this paper ex-plores employer experiences with CDHP in two areas. First, it describesthose firms most likely to offer CDHP. Second, this paper estimates poten-tial cost reductions from an emphasis on CDHP. Results indicate large,non-union employers offer CDHP most often. Further, an emphasis onCDHP plans is not associated with a reduction in average contributionsto insurance policies.
∗Western Kentucky University
1 Introduction
Employer financing of health benefits creates a unique relationship between
employers and employees. When employers guarantee defined benefits, rising
health care costs impose an upward pressure on total employee compensation.
Thus, the potential benefits of cost-sharing, coupled with preferential treatment
by the tax code, bring consumer driven health plans (CDHP) sharply into focus
for firms seeking relief from expensive insurance premiums. This paper ana-
lyzes employer experiences with CDHP, describing firms which offer CDHP and
estimating whether offering CDHP influences expenditures on health benefits.
Health care expenditures in the U.S., measured as a proportion of GDP,
rapidly increased throughout the previous decade. The pace of this growth,
which peaked in 2002 at a rate of 9.1%, slowed through 2005 and 2006. However,
projections indicate health care spending will increase at a rate faster than GDP
growth for the foreseeable future, potentially accounting for 20% of GDP by 2015
(Borger et al., 2006).
Rising aggregate expenditures translate into higher insurance premiums,
raising concerns over equity and access to care. High insurance premiums dis-
courage employers from expanding coverage, increasing the proportion of Amer-
icans lacking health insurance (Gabel et al., 2004), and encouraging employers
to seek alternative methods of financing health benefits (Gabel et al., 2003).
In this context, consumer directed health plans could quickly become a central
part of the American health finance system.
Understanding the implications CDHP expansion could exact on employers
1
merits attention. Initially, by influencing the mix of goods included in compen-
sation packages, such expansion would alter the nature of the employer-employee
relationship. Second, because many employers bemoan rising health care costs
as an impediment to international competitiveness, CDHP expansion could po-
tentially impact the strength of the US economy as a whole (Reinhardt et al.,
2004). Finally, employers prioritize the health of their workforce because health
status impacts productivity in the workplace. Short term absenteeism repre-
sents as much as 20% of all health related costs (Goetzel et al., 2003).
This paper focuses on employers who offer consumer driven health plans.
Utilizing data from the Kaiser-HRET Employer Health Benefits Annual Survey,
the present analysis describes the type of firm most likely to adopt CDHP as
a cost containment mechanism. Results indicate small firms offer CDHP less
often than medium and large firms. Further, this paper attempts to gauge the
effectiveness of CDHP plans regarding controlling expenditures on health care
relative to traditional insurance offerings. Results suggest CDHP failed to reign
in expenditures on health benefits.
2 CDHP History & Experience
Advocates of cost sharing, beginning with Feldstein (1973), blame the presence
of third party payers for increasing health care costs. By shielding consumers
from the marginal cost of treatment, conventional insurance policies result in
excess demand. Consumer driven health plans rectify this problem by increasing
the marginal cost of service faced by consumers. Advocates of CDHP argue
2
thrifty consumers will impose discipline on the health care market by demanding
information regarding price and quality. In this view, competition on price and
quality holds the key to improving health outcomes and access to care while still
reducing prices (see Cannon (2006), Buntin et al. (2005), and Hughes-Cromwick
et al. (2007), among others).
The original incarnation of the consumer directed health ideology, the Archer
Medical Savings Account (MSA), failed to attract significant enrollment (Davis
et al., 2005). The Medicare Modernization Act of 2003 rolled existing MSAs into
Health Savings Accounts (HSAs). The act, among other goals, put forth the
HSA, coupled with a high deductible health plan (HDHP), as an alternative to
traditional insurance models. Thus, the Medicare Modernization Act represents
the boldest move, to date, toward integrating CDHP into the health finance
system.
Proponents of HSAs isolated small businesses as fertile territory for the
growth of CDHP. Members of the Bush administration, among others, sug-
gested HSAs as a low cost alternative to conventional health care for small
business owners, potentially extending health insurance to a class of previously
neglected workers. Early evidence indicated strength in this area. As of 2004,
among employers purchasing small group HSA coverage 16% covered previously
uninsured individuals, a further 30% of HSA policies sold directly to individuals
covered previously uninsured customers (Chovan and Yoo, 2004).
However, the critical barrier to widespread expansion of insurance coverage
remains the decision to offer insurance on the part of small businesses. Small
3
businesses base insurance offering decisions on a number of factors beyond cost.
Factors such as worker demand, labor market competition and labor force com-
position generally take precedence over price in such determinations (Hadley
and Reschovsky, 2002). In fact, probability estimations by Gates et al. (2008),
based on an earlier iteration of the Kaiser-HRET survey, confirm that small
firms do not offer HSAs at higher rates than other firms. Further, Gates et al.
(2008) report a non-linear relationship between firm size and CDHP offering be-
havior, whereby medium sized firms (200-499 employees) offer CDHP at lower
rates than either smaller or larger firms.
Much of the literature to date focuses on the consumer-level enrollment deci-
sion (for example, Cardon and Showalter (2001)) and the potential system-wide
impacts of a paradigmatic shift toward CDHP (for example, Davis (2004)).
Thus, the marginal impact of CDHP adoption on employer health benefits
spending remains an open question.
Many employers report favorable experience with CDHP (Prince, 2003).
However, anecdotal reports fail to describe the broader firm-level experience
with health savings accounts. Nor do they incorporate broader macroeconomic
influences that impact costs and expenditures (Buntin et al., 2006).
Further, achieving a reduction in firm expenditures on health benefits re-
quires that CDHPs significantly increase cost sharing, explicitly shifting the fi-
nancial burden onto employees. Thus, CDHPs increase cost sharing only when
they replace a more generous insurance type. Many conventional (non-CDHP)
insurance plans already incorporate significant cost sharing mechanisms. In
4
fact, for most high-spending consumers enrollment in an HSA/HDHP combina-
tion would actually represent a decrease in cost sharing, rather than an increase
(Remler and Glied, 2006).
This paper adds to the current literature in two areas. First, this paper seeks
to replicate and verify the results presented in Gates et al. (2008) by estimat-
ing the relationship between firm size and CDHP offering. Second, this paper
attempts to ascertain whether or not firms that emphasize CDHP experience
cost reductions.
3 Data & Empirical Methodology
3.1 Data
The present analysis utilizes data from the 2007 Kaiser Health Research Educa-
tion Trust Survey on Employee Benefits. The Kaiser-HRET survey, conducted
from January to May 2007, describes the almost 2000 firms that completed the
entire survey, representing an overall response rate of 49%(Claxton et al., 2007).
The survey describes health benefits in depth, including plans offered, plan
design, cost and annual firm contributions. Regarding those firms which do not
offer consumer directed plans, the survey measures employer attitudes toward
offering these plans in the future.
Additionally, the survey includes descriptive data on the firm and its em-
ployees, including firm size, union presence, et cetera. The Kaiser-HRET survey
provides a significant amount of depth on each of these areas, making it uniquely
suited to provide insight into the questions at hand.
5
3.2 Firm Offering Behavior
The present analysis first examines the influence firm characteristics and work-
force demographics exert on CDHP offering behavior. The analysis considers
two estimations based on the equation;
yf = cf + β1size′f + β2industry′
f + β3region′f + β4workforce′
f + εf . (1)
The first of these estimations isolates firms that already offer an HSA and/or
an HRA. Further, the Kaiser-HRET survey asks employers about their plans to
offer CDHP. The second equation estimates the probability of a firm responding
it was “very likely” or “somewhat likely” to offer CDHP.
Model specifications loosely follow those used by Gates et al. (2008). By
utilizing the 2007 iteration of the Kaiser-HRET survey, the results presented
below should compliment and update Gates et al. (2008), who only had access
to the 2003-2006 versions of the survey. Controls include firm size1, region and
workforce demographics.
However, Gates et al. (2008) include continuous measurements of percentage
of low income and part time workers, assuming a linear relationship between
workforce demographics and offering behavior. The estimations above utilize a
set of dichotomous indicators, isolating workforces with at least 35% low income
workers, at least 35% high income workers and at least 35% full time workers2.
These binary controls focus the estimations on differences between categories of1The probability estimates measure firm size using six dichotomous categories, while the
Tobit estimations later divide firm size into three categories.2Marginal effects associated with a one-percent change in continuous measurements of
income and hours worked were qualitatively similar, but the scale of the effect hinders inter-pretation.
6
firms, rather than the marginal effect associated with small changes in workforce
composition.
The calculation of marginal effects simplifies the interpretation of probit coef-
ficients. Marginal effect estimations generally consider each variable separately,
holding all other variables at their means. However, similar to the methodol-
ogy employed by Boonen et al. (2008), the marginal effects presented in table
2 result from averaging observation level marginal effects. Marginal effects are
calculated as derivatives for continuous variables, and as discrete changes for
binary variables. Standard errors are calculated a via Monte-Carlo simulation
with 500 replications.
3.3 CDHP & Average Contributions
Second, this paper estimates the influence CDHP offering exerts on firm contri-
butions to health benefits. These equations estimate average firm contributions,
per policy, separately for family and individual policies. The presence of missing
and zero values for average contributions requires censoring dependent variables
at 0, using Tobit estimations.
Firm-specific controls (represented by X′f , below) include firm size, industry,
and region. Further, these estimations include the percentage of employees
covered by insurance and the percentage by which cost increased over last year,
attempting to control for influences on price exogenous to benefit design.
Two separate model designs estimate the influence of CDHP offering on
7
average contributions. First, the equation
yf = cf + β1X′f + β2plan′
pf + εf (2)
utilizes a series of binary variables, represented by the vector plan′f , indicating
whether firm f offers a given plan type, including HMO, PPO, HRA, and HSA
plans. The reference group is conventional insurance policies (POS). Among
the variables represented by the vector X′f , these estimations control for the
number of insurance varieties, purging from the insurance variables any effect
associated with a diverse portfolio of options.
Second, The Kaiser-HRET survey includes data for overall average contri-
butions, but also includes average contributions to each type of policy. Contri-
butions to each type of plan p are included continuously, as proportions of total
spending3 in the equation
yf = cf + β1X′f + β2
contributionpf
totalf+ εf . (3)
These proportional variables measure the emphasis firms place on each type of
plan, distinguishing firms who simply experiment with CDHP from those who
heavily rely on it.
If, as suggested by Gabel et al. (2002), firms utilize CDHP introduction to de-
crease health care expenditures, firm specific heterogeneity could bias estimates.
Fully disambiguating the relationship between benefit design and average contri-
butions requires purging this heterogeneity. A variety of estimation techniques,
including an instrumental variable estimation, failed to produce significant re-3For ease of interpretation, these variables are also multiplied by 100, such that these
variables are measured in whole units instead of hundredths.
8
sults. The specifications above correct for this problem via the inclusion of a
host of controls. Controlling for firm size, labor force characteristics (especially
income), and percent cost increases should capture variation associated only
with high spending.
4 Results
Overall, results challenge the expectations that many have for CDHP financ-
ing vehicles. HRAs and HSAs remain unpopular, especially among small firms.
Additionally, although firms introduce CDHP as a cost containment mecha-
nism, such behavior lacks an association with lower average spending. Tables
below present sample means and results from the estimations described in the
preceding section.
4.1 Firm Offering Behavior
Results of CDHP firm offering behavior estimations appear in table 2. Control
variables return unsurprising results. Firms with low income workforces, at
least 35% of employees earning less than $21,000, are 5.6 percentage points less
likely to offer either an HSA or an HRA, and 5.2 percentage points less likely
to indicate a willingness to offer CDHP in the future. Some industries display
an affection for CDHP. For example, financial firms are 15 percentage points
more likely to offer CDHP, and 16 percentage points more likely to indicate a
willingness to offer CDHP in the future.
The most striking result from the probability estimations regard the rela-
tionship between firm size and CDHP offering. HRAs and HSAs remain rare,
9
offered by only 10% and 5% of firms4, but these policies become more rare as
firm size decreases.
Firms with 50 to 199 employees have the lowest statistically significant prob-
ability of offering CDHP relative to the reference group (firms with 3 to 24
employees)5. The probability steadily increases, to 12 percentage points for
firms with 1000 to 4999 employees. The largest category of firms, those with
more than 5000 employees, offers CDHP at a significantly higher rate than any
other group. Despite these differences in actual offering behavior, the rate at
which benefit managers report a willingness to offer CDHP in the future seems
unrelated to firm size.
4.2 CDHP & Average Contributions
Estimations of average contributions to family and single insurance policies
provide robust results, presented in table 3. However, the two specifications of
these models produce parameter estimates of differing magnitudes. As indicated
by their corresponding σ (sigma) statistics, the second specification returns
slightly higher standard errors than the first. Also, Tobit estimations occur in
two steps, limiting the interpretation of the coefficients presented and discussed
below to the uncensored portion of the dependent variable (Roncek, 1992) 6.
Meaning, results apply only to firms with nonzero average contributions to
insurance policies.
As suggested by Hadley and Reschovsky (2002), workforce characteristics
4For sample means, see table 1.5The smallest category of firms, 25 to 49 employees, differs insignificantly from the reference6However, in this case the Tobit estimator censors a small portion of the overall data set.
OLS results, not presented, based on the same specifications return similar estimations.
10
exert a strong influence on health care expenditures. Union presence increases
average contributions by at least $1600 for family plans and at least $150 for sin-
gle plans. Results also quantify the differential between health benefits for high
income and low income workers. Firms with high concentrations of low income
workers contribute as much as $963 less on average to family plans. Firms with
similarly high concentrations of high income workers contribute an average of
$1400 more according to the first specification, and $2300 according to the sec-
ond. Finally, firms in the Northeast spend significantly more than firms in other
regions. The two specifications return parameters of similar magnitude, indi-
cating firms located in the Northeast contribute $3100 more per family policy
and at least $350 more per single policy.
The variables of interest in these models measure the influence insurance
offerings exert on average contributions. Traditional point of service plans serve
as the reference group in each of these estimations. The first specification reports
an association between CDHP offering and an increased average contributions
to both family and single plans, $10,000 and $4,000 respectively. Similarly, firms
which offer an HRA contribute $8,400 more to family plans and $3,500 more to
single plans.
The second specification tells a similar story. The coefficients associated with
the proportional variables indicate firms which dedicate 1% of total spending
on HSAs contribute $422 more per family plan and $165 more per single plan.
This estimation implies that a firm which dedicates 10% of total spending to an
HSA plan would spend, on average, $4,220 more.
11
5 Discussion
The results presented above give pause, especially in light of the assertions made
by CDHP advocates. Employer experiences with CDHP indicate expansion of
coverage associated with CDHP may be limited. Small firms offer CDHP pro-
grams significantly less often than other firms. The estimations also indicate
firm attitudes regarding CDHP are unrelated to firm size. These findings ques-
tion the ability of CDHP plans to extend coverage to traditionally underserved
populations.
Advocates argue HSAs, due to their low cost, could expand coverage among
low income workers (Baicker et al., 2007). However, the lack of CDHP offering
on the part of employers hinders the effectiveness of HSAs in this regard. Each
of the probability estimates describe firms with low income workforces as sub-
stantially less likely to engage in CDHP offering behavior, echoing the finding
reported by Parente et al. (2004) that HSAs primarily serve wealthy individuals.
The probability estimates also sketch the potential for CDHP growth. The
third probability estimation isolates firms in financial, wholesale and transporta-
tion industries as most likely to offer CDHP in the future. Despite the variety of
controls included in the probability estimations, industry-wide preferences for
CDHP adoption remain unexplained. Taken together, however, the estimations
suggest stability within CDHP offering trends. Expected growth will occur in
the same sectors, among firms similar to those who currently offer CDHP.
Given the firm specific heterogeneity discussed above, interpreting the aver-
age contribution estimation results requires a degree of agnosticism regarding
12
causality. Results clearly indicate an association between CDHP offering and
average contributions by which firms offering CDHP contribute inordinately
more per policy than other firms. Results support at least two interpretations
of this association. First, the association could simply imply an association
between high firm contributions and a desire for cost containment.
Second, CDHP policies could operate in a way that advocates did not foresee,
serving as a compliment to traditional insurance policies, rather than a substi-
tute. Tax-preferential treatment of contributions to medical savings vehicles
could make them an efficient way for employers to increase total compensation.
CDHP policies offered strictly as an addition to traditional insurance options
must increase average contributions. The results associated with workforce char-
acteristics buttress this hypothesis.
Although informative, the results above highlight the need for a more thor-
ough understanding of the relationship between benefit package design and
average contributions. The method by which consumers finance health care
purchases heavily exerts system wide effects on competitive and strategic inter-
actions between providers. In this context, the role employers play as a de facto
financial intermediary between insurance providers and employees makes firms
a key component of systematic reform.
By investigating the employer experience, this paper attempts to gauge the
success of the consumer driven health plan movement. In sum, results suggest
skepticism. The lack of CDHP adoption by small firms, coupled with an asso-
ciation between high spending and CDHP offering does not seem to function as
13
an effective cost control mechanism. Consumer directed health plans may hold
unrealized potential. However, as currently implemented HSAs and HRAs do
not represent revolutionary approaches to cost containment.
14
References
Baicker, K., Dow, W. H., and Wolfson, J. (2007). Lowering the barriers to
consumer directed health care: Responding to concerns. Health Affairs,
26(5):1328–1332.
Boonen, L., Schut, F., and Koolman, X. (2008). Consumer channeling by health
insurers: natural experiments with preferred providers in the Dutch pharmacy
market. Health Economics, 17(3):299–316.
Borger, C., Smith, S., Truffer, C., Keehan, S., Sisko, A., Poisal, J., and Clemens,
K. M. (2006). Health spending projections through 2015: Changes on the
horizon. Health Affairs, 25(2):w61–73.
Buntin, M. B., Damberg, C., Haviland, A., Kapur, K., Lurie, N., Mcdevitt, R.,
and Marquis, M. S. (2006). Consumer-Directed Health Care: Early Evidence
About Effects On Cost And Quality. Health Affairs, 25(6):516–530.
Buntin, M. B., Damberg, C., Haviland, A., Lurie, N., Kapur, K., and Marquis,
S. M. (2005). Consumer-directed Health Plans: Implications for Health Care
Quality and Cost. Rand Corporation.
Cannon, M. (2006). Health savings accounts: Do the critics have a point? Cato
Policy Analysis, (569).
Cardon, J. H. and Showalter, M. H. (2001). An examination of flexible spending
accounts. Journal of Health Economics, 20(6):935–954.
15
Chovan, T. and Yoo, H. (2004). Health savings accounts off to a fast start in
the individual market. America’s Health Insurance Plans (AHIP).
Claxton, G., Dijulio, B., Finder, B., Becker, E., Hawkins, S., Pickreign, J.,
Whitmore, H., and Gabel, J. (2007). Employer Health Benefits: 2007 Annual
Survey. Henry J. Kaiser Family Foundation; Health Research and Educational
Trust.
Davis, K. (2004). Consumer-directed health care: Will it improve health system
performance? Health Services Research, 39(4p2):1219–1234.
Davis, K., Doty, M., and Ho, A. (2005). How High Is Too High? Implications
of High-Deductible Health Plans. The Commonwealth Fund, April.
Feldstein, M. S. (1973). The welfare loss of excess health insurance. The Journal
of Political Economy, 81(2):257–280.
Gabel, J., Claxton, G., Gil, I., Pickreign, J., Whitmore, H., Holve, E., Finder,
B., Hawkins, S., and Rowland, D. (2004). Health benefits in 2004: Four years
of double-digit premium increases take their toll on coverage. Health Affairs,
23(5):200–209.
Gabel, J., Claxton, G., Holve, E., Pickreign, J., Whitmore, H., Dhont, K.,
Hawkins, S., and Rowland, D. (2003). Health benefits in 2003: Premiums
reach thirteen-year high as employers adopt new forms of cost sharing. Health
Affairs, 22(5):117–126.
16
Gabel, J. R., Sasso, A. T. L., and Rice, T. (2002). Consumer-Driven Health
Plans: Are They More Than Talk Now? Health Affairs, 20:49–64.
Gates, S. M., Kapur, K., and Mandic, K. P. (2008). Consumer-Directed Health
Plans and Health Savings Accounts: Have They Worked for Small Businesses?
Forum for Health Economics and Policy, 11(2):4.
Goetzel, R., Hawkins, K., Ozminkowski, R., and Wang, S. (2003). The Health
and Productivity Cost Burden of the” Top 10” Physical and Mental Health
Conditions Affecting Six Large US Employers in 1999. Journal of Occupa-
tional and Environmental Medicine, 45(1):5.
Hadley, J. and Reschovsky, J. D. (2002). Small Firms’ Demand for Health
Insurance: The Decision to Offer Insurance. Inquiry, 39(2):118–137.
Hughes-Cromwick, P., Root, S., and Roehrig, C. (2007). Consumer-driven
healthcare: information, incentives, enrollment, and implications for national
health expenditures; there are reasons for optimism. Business Economics.
Parente, S. T., Feldman, R., and Christianson, J. B. (2004). Employee Choice
of Consumer-Driven Health Insurance in a Multiplan, Multiproduct Setting.
Health Services Research, 39(4p2):1091–1112.
Prince, M. (2003). Consumer-driven health plans seeing some initial successes.
Business Insurance, 37(13):p3 –.
Reinhardt, U. E., Hussey, P. S., and Anderson, G. F. (2004). U.s. health care
spending in an international context. Health Affairs, 23(3):10–25.
17
Remler, D. K. and Glied, S. A. (2006). How much more cost sharing will health
savings accounts bring? Health Affairs, 25(4):p1070 – 1078.
Roncek, D. (1992). Learning more from tobit coefficients: Extending a compara-
tive analysis of political protest. American Sociological Review, 57(4):503–507.
18
Table 1: Sample MeansN = 1839
Min Max Mean Std. Dev
Firm Size 3 to 24 0 1 0.116 0.320
(6 Categories) 25 to 49 0 1 0.074 0.263
50 to 199 0 1 0.144 0.351
200 to 999 0 1 0.245 0.430
1000 to 4999 0 1 0.247 0.431
5000 and above 0 1 0.175 0.380
Industry Service 0 1 0.368 0.482
Mining 0 1 0.009 0.096
Construction 0 1 0.038 0.190
Transportation / Utility 0 1 0.049 0.216
Wholesale 0 1 0.055 0.229
Financial 0 1 0.074 0.263
Government 0 1 0.368 0.482
Health Care 0 1 0.091 0.287
Retail 0 1 0.063 0.242
Manufacturing 0 1 0.139 0.346
Region South 0 1 0.322 0.468
Northeast 0 1 0.211 0.408
Midwest 0 1 0.290 0.454
West 0 1 0.177 0.382
Workforce Description Urban 0 1 0.817 0.387
Union 0 1 0.326 0.469
Full Time 0 1 0.056 0.230
Low Income 0 1 0.151 0.358
High Income 0 1 0.382 0.486
% Covered 0.04 100 0.667 0.240
Cost Increases Conventional 0 50 0.485 2.908
HMO 0 81 2.793 6.036
PPO 0 58 4.934 6.962
POS 0 48 1.719 4.752
Plans Offered Number of Choices 1 5 1.522 0.711
HSA 0 1 0.049 0.217
HRA 0 1 0.096 0.295
HMO 0 1 0.325 0.468
PPO 0 1 0.744 0.437
Proportion of Spending byType (Family)
HMO 0 100 16.518 27.538
PPO 0 100 55.119 41.080
HRA 0 50 1.678 7.969
HSA 0 50 2.944 9.688
Proportion of Spending byType (Single)
HMO 0 100 16.787 27.726
PPO 0 100 55.622 40.915
HRA 0 50 1.804 8.306
HSA 0 50 3.002 9.798
Tab
le2:
Pro
babi
lity
Est
imat
es(M
argi
nal
Effe
cts)
:F
irm
CD
HP
Offe
ring
Beh
avio
r
Offe
rsH
RA
and/
orH
SAL
ikel
yto
Ado
ptC
DH
PV
ari
ab
leM
arg
inal
Eff
ect
Std
Err
Marg
inal
Eff
ect
Std
Err
Fir
mS
ize
(No.
of
Em
plo
yees)
25
to49
0.0
778
0.0
498
−0.0
196
0.0
435
50
to199
0.1
056
0.0
435**
0.0
322
0.0
396
200
to999
0.0
847
0.0
382**
0.0
439
0.0
363
1000
to4999
0.1
215
0.0
424**
0.0
554
0.0
398
5000
an
dab
ove
0.1
916
0.0
470**
0.0
620
0.0
410
Ind
ust
ry
Min
ing
0.1
651
0.1
313
0.0
689
0.1
193
Con
stru
ctio
n0.0
851
0.0
611
0.0
382
0.0
605
Tra
nsp
ort
ati
on
/U
tility
0.0
902
0.0
594
0.1
320
0.0
627**
Wh
ole
sale
0.1
225
0.0
584**
0.1
338
0.0
592**
Fin
an
cial
0.1
587
0.0
593**
0.1
609
0.0
591**
Gover
nm
ent
0.0
570
0.0
349
0.0
118
0.0
369
Hea
lth
Care
0.0
430
0.0
458
0.0
554
0.0
493
Ret
ail
0.0
717
0.0
518
0.1
332
0.0
547**
Manu
fact
uri
ng
0.0
880
0.0
446**
0.0
706
0.0
448
Regio
nN
ort
hea
st−
0.0
208
0.0
232
−0.0
048
0.0
284
Mid
wes
t0.0
373
0.0
220*
0.0
326
0.0
261
Wes
t0.0
022
0.0
233
−0.0
245
0.0
268
Workfo
rce
Desc
rip
tion
Urb
an
−0.0
150
0.0
236
−0.0
222
0.0
279
Un
ion
Pre
sen
ce−
0.0
322
0.0
179*
−0.0
570
0.0
222**
Fu
llT
ime
−0.0
567
0.0
303*
0.0
004
0.0
429
Low
Inco
me
−0.0
560
0.0
197**
−0.0
526
0.0
262**
Hig
hIn
com
e0.0
083
0.0
175
0.0
263
0.0
220
Sig
nifi
can
cele
vel
s,(*
)=
10%
(**)
=5%
20
Tab
le3:
Tob
itE
stim
atio
ns(C
enso
red
atZ
ero)
:C
ontr
ibut
ions
toIn
sura
nce
Pol
icie
s
Fam
ily
Pla
nS
ingle
Pla
nF
am
ily
Pla
nS
ingle
Pla
nV
ari
ab
leE
stim
ate
Std
Err
Est
imate
Std
Err
Est
imate
Std
Err
Est
imate
Std
Err
Inte
rcep
t-4
855.3
65
822.9
36
**
-758.5
24
298.1
02
**
2206.8
35
394.7
50
**
2354.1
10
397.6
65
**
Fir
mS
ize
Sm
all
Fir
m(3
to199)
-1806.6
00
393.6
92
**
520.5
54
142.7
48
**
-5376.4
17
450.2
46
**
-1022.5
56
178.1
54
**
Med
ium
Fir
m(2
00
to999)
464.9
0377.7
98
694.1
1137.1
97
**
−1359.1
7451.0
67
**
−72.1
3177.2
10
Ind
ust
ry
Ind
ust
rial
−588.6
0382.7
41
−1172.9
0138.7
31
**
−715.7
3474.9
45
−1234.7
5181.2
85
**
Sale
s−
758.2
2407.6
77
−934.4
8147.7
53
**
−761.9
8505.3
11
−975.9
9193.0
91
**
Hea
lth
&G
over
nm
ent
1480.0
0405.0
25
**
332.3
9146.8
60
*830.8
0500.1
52
76.3
5191.7
95
Regio
nN
ort
hea
st3136.6
2411.8
23
**
349.4
0149.3
03
*3196.9
0507.0
39
**
398.8
6195.5
63
*M
idw
est
1358.2
5376.3
23
**
146.7
3136.3
71
424.0
9459.4
36
−198.6
6177.7
73
Wes
t637.3
6432.5
98
286.5
8156.8
07
22.8
5530.6
25
42.3
8204.9
52
Workfo
rce
Desc
rip
tion
Urb
an
389.1
9397.1
64
−100.8
3143.7
22
1215.3
5467.0
17
**
239.7
9187.5
17
Un
ion
pre
sen
ce1681.6
2342.9
82
**
779.8
1124.4
47
**
2031.1
6424.3
45
**
922.3
2162.4
64
**
Low
Inco
me
−963.8
3422.7
52
*−
313.9
4153.0
95
*−
1677.2
8501.8
53
**
−620.4
3200.3
73
**
Hig
hIn
com
e1474.7
0317.1
55
**
310.0
6115.0
16
**
2319.7
4392.5
21
**
667.6
4149.8
54
**
Per
cent
Fu
llT
ime
1110.6
9675.7
06
151.2
5244.8
71
910.7
2724.5
91
119.3
8320.2
92
Per
cent
Em
plo
yee
sC
over
ed3670.2
6683.4
97
**
1159.5
2247.5
37
**
5594.9
9551.0
27
**
1974.6
3322.5
50
**
Cost
Increase
sC
onven
tion
al
−275.6
455.0
06
**
−148.6
919.9
17
**
−56.6
562.7
71
−54.0
223.9
61
*H
MO
−14.2
233.0
42
−13.2
511.9
90
202.0
939.1
62
**
78.7
814.8
80
**
PP
O34.9
023.5
76
15.4
48.5
55
150.0
628.3
39
**
62.2
810.8
55
**
PO
S271.2
937.2
74
**
89.7
513.5
07
**
482.3
943.0
68
**
176.7
717.1
00
**
Insu
ran
ce
Off
erin
gs
Nu
mb
erof
Pla
nC
hoic
es5534.7
4418.6
03
**
2401.9
8151.6
58
**
HS
A10760.6
2741.2
54
**
3999.8
2268.3
55
**
HR
A8427.9
6625.9
25
**
3484.0
8226.8
47
**
HM
O3548.3
0593.2
08
**
1314.3
7214.8
20
**
PP
O3394.6
5491.9
54
**
1111.3
2177.7
87
**
Prop
orti
on
of
Sp
en
din
gD
ed
icate
dto
Each
Typ
eof
Poli
cy
(Fam
ily)
HM
O77.1
63
9.6
11
**
PP
O26.6
81
6.4
14
**
HR
A351.9
87
24.2
54
**
HS
A422.3
37
20.5
26
**
Prop
orti
on
of
Sp
en
din
gD
ed
icate
dto
Each
Typ
eof
Poli
cy
(Sin
gle
)
HM
O27.8
28
3.7
32
**
PP
O7.2
90
2.6
34
**
HR
A122.9
75
9.1
31
**
HS
A165.6
89
8.0
42
**
σ6072.4
20
102.4
23
**
2205.8
49
36.8
77
**
7539.9
20
126.9
77
**
2884.8
11
48.2
25
**
Sig
nifi
can
cele
vel
s,(*
)=
10%
(**)
=5%